Downregulation of rRNA synthesis by BCL-2 induces chemoresistance in diffuse large B cell lymphoma

Abstract

Overexpression of the antiapoptotic oncogene BCL-2 predicts poor prognosis in diffuse large B cell lymphoma (DLBCL) treated with anthracycline-based chemoimmunotherapy. Anthracyclines exert antitumor effects by multiple mechanisms including inhibition of ribosome biogenesis (RiBi) through rRNA synthesis blockade. RiBi inhibitors induce p53 stabilization through the ribosomal proteins-MDM2-p53 pathway, with stabilized p53 levels depending on baseline rRNA synthesis rate. We found that the BH3-mimetic venetoclax could not fully reverse BCL-2-mediated resistance to RiBi inhibitors in DLBCL cells. BCL-2 overexpression was associated with decreased baseline rRNA synthesis rate, attenuating p53 stabilization by RiBi inhibitors. Drugs stabilizing p53 irrespective of RiBi inhibition reversed BCL-2-induced resistance in vitro and in vivo, restoring p53 activation and apoptosis. A small nucleolar size, indicative of low baseline rRNA synthesis, correlated with high BCL-2 levels and poor outcomes in DLBCL patients. These findings uncover alternative BCL-2-dependent chemoresistance mechanisms, providing a rationale for specific combination strategies in BCL-2 positive lymphomas.

Keywords: Cancer; Cell biology.

Graphical abstract

Figure 1

BCL-2 overexpression promotes resistance to inhibition of ribosome biogenesis in diffuse large B cell lymphoma cell lines (A) Bar graph showing antiproliferative effects of CHOP chemotherapy on 12 diffuse large B cell lymphoma (DLBCL) cell lines, measured by CellTiter-Glo assay (CTG). Cells were incubated for 24 h with vincristine 0.37 nM, doxorubicin 50 nM, acrolein 1.5 μM, and methylprednisolone 12.5 μM. Error bars represent standard deviation (SD) of triplicate experiments (n = 3). The heatmap shown below indicates the TP53, MYC and BCL-2 status and the cell of origin (COO). (B) CTG assay showing the effects of increasing doses of doxorubicin and actinomycin D on cell viability in 12 DLBCL cell lines after 24 h: in gray, TP53 mutant cell lines; in blue, BCL-2 positive/TP53 WT cell lines; in red, the BCL-2 negative/TP53 WT SUDHL-5 cell line. Error bars represent SD of triplicate experiments (n = 3). (C) qPCR analysis of 45S rRNA expression in three DLBCL cell lines incubated for 6 h with the indicated treatments: CHOP (vincristine 0.75 nM, doxorubicin 100 nM, acrolein 3 μM, and methylprednisolone 25 μM) and single agents RiBi inhibitors. Error bars represent SD of five independent experiments (n = 5). Student’s t test: ∗∗p < 0.01, ∗∗∗p < 0.005. (D) Scatterplot representing the effects of 12 h incubation with CHOP and single agent RiBi inhibitors on caspase 3/7 activation (as measured by Caspase-Glo assay) in SUDHL-5, TMD8, and SUDHL-6 cell lines. Each dot represents the mean of triplicate experiments (n = 3). Student’s t test: ∗p < 0.05. (E) Scheme of BCL-2 overexpression experiments. Western blot shows BCL-2 protein levels in SUDHL-5 cells carrying the empty vector (empty) or the BCL-2 TET-ON inducible system (BCL-2) after 96 h of incubation with 1 μg/ml doxycycline. (F) CTG assay showing the cytotoxic effects at 24 h of three doses of CHOP (vincristine 1.5-0.75-0.37 nM, doxorubicin 200-100-50 nM, acrolein 6-3-1.5 μM, and methylprednisolone 50-25-12.5 μM) in SUDHL-5 cells in the presence or absence of BCL-2. Error bars represent SD of five independent experiments (n = 5). Student’s t test: ∗∗∗p < 0.005. (G) Caspase-Glo assay showing levels of caspase 3/7 activation in SUDHL-5 cells treated for 12 h with CHOP (vincristine 0.75 nM, doxorubicin 100 nM, acrolein 3 μM, and methylprednisolone 25 μM) in the presence or absence of BCL-2. Error bars represent SD of triplicate experiments (n = 3). Student’s t test: ∗p < 0.05. (H and I) Effects of 24 h treatment with single agent RiBi inhibitors on cell viability (H) and apoptosis (I) in the presence or absence of BCL-2. Error bars represent SD of five independent experiments (n = 5). Student’s t test: ∗∗p < 0.01, ∗∗∗p < 0.005.

Figure 2

Enforced BCL-2 expression is associated with a reduced rRNA synthesis rate in DLBCL cell lines (A) Bar graph showing the cytotoxic effects of the indicated treatments in the absence or presence of BCL-2. Cells were preincubated with doxycycline 1 μg/ml for 96 h and then treated for 24 h. Error bars represent standard deviation (SD) of five experiments. Student’s t test: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.005. V, venetoclax; D, doxorubicin; A, actinomycin D; and C, CX-5461. (B) qPCR analysis showing the levels of BCL-2 mRNA and 45S rRNA in SUDHL-5 cells transduced with an Empty vector or a BCL-2 doxycycline (TET-ON) inducible system, treated with doxycycline for 96 h. Error bars represent SD of five independent experiments (n = 5). Student’s t test: ∗∗p < 0.01. (C) Co-immunoprecipitation (coIP) experiment showing the effects of BCL-2 overexpression on MDM2-p53-RPs binding in SUDHL-5 cells in the presence or absence of BCL-2. Cells were incubated with doxycycline for 96 h before harvesting for coIP analysis. Left: western blot analyses showing the expression levels of the indicated proteins in the initial lysates (INPUT) and the levels of MDM2-P53 or MDM2-RPs complexes immunoprecipitated with an MDM2 antibody. Right: protein quantification of the coIP experiment performed using the ImageJ software. (D) Immunoprecipitation-qPCR (IP-qPCR) analysis demonstrating the interaction between nucleolin and BCL-2 mRNA in SUDHL-5 cells. The experiment was conducted in the presence or absence of the BCL-2 coding sequence (BCL2) or the BCL2 3′ untranslated region (B-UTR). Cells were treated with doxycycline for 96 h prior to nucleolin immunoprecipitation. qPCR analysis quantified the BCL2 mRNA levels in the initial sample (INPUT) and in the nucleolin-BCL2 mRNA complexes immunoprecipitated using a nucleolin-specific antibody (IP nucleolin). Error bars represent SD of triplicate experiments (n = 3). Student’s t test: ∗∗∗p < 0.005. (E) Representative immunofluorescence images of SUDHL-5 cells showing the expression patterns of nucleolin (red) and BCL-2 (green) proteins. Cells were transduced with an empty vector (empty), BCL-2 (BCL-2), or the BCL-2 3′ untranslated region (B-UTR) using a TET-ON system and were incubated with doxycycline for 96 h. (F) Representative western blots showing total protein expression levels of nucleolin and BCL-2, under the same experimental conditions as described in Figure 2E. (G) Immunohistochemistry representative images showing three different patterns of nucleolin distribution in primary FFPE DLBCL tissues, depending on BCL-2 expression levels. Bar, 10 μm. (H) Violin plot showing the correlation between the three patterns of nucleolin distribution and BCL-2 mRNA levels, as measured by T-GEP. Student’s t test: ∗p < 0.05, ∗∗p < 0.01.

Figure 3

BCL-2 overexpression attenuates p53 stabilization and activation following RiBi inhibitors treatment (A) qPCR analysis showing the expression levels of 45S rRNA and TP53 mRNA in SUDHL-5 cells incubated for 24 h with the indicated treatments in the presence or absence of BCL-2. Error bars represent standard deviation (SD) of triplicate experiments (n = 3). Student’s t test: ∗p < 0.05,∗∗p < 0.01. (B) Representative immunoblots showing the effect of single agents RiBi inhibitors on P53 protein levels in SUDHL-5 cells in the presence or absence of BCL-2 (same conditions as Figure 3A). Protein quantification (right) was performed using ImageJ software. (C) RNA-seq experiment performed in SUDHL-5 cells incubated with RiBi inhibitors for 6 h in the presence or absence of BCL-2 (after 96 h induction with doxycycline). The experiment was performed in triplicate (n = 3). Differentially expressed genes (DEGs) were reported in the Venn diagrams. A consensus DEGs list was extracted and matched with the TRANSFAC database to focus on differentially regulated p53 target genes and explore their expression in the absence or presence of BCL-2 following treatment with RiBi inhibitors. (D) Radar chart displaying fold changes of significantly regulated p53 target genes identified in SUDHL-5 cells treated with actinomycin D in the presence (blue line) and absence (black line) of BCL-2. (E) Bar graph showing fold change of common p53 target genes identified in SUDHL-5 cells following treatment with different RiBi inhibitors in the presence or absence of BCL-2. (F) qPCR analysis showing expression of representative p53 targets, CDKN1A (p21) and PUMA, in SUDHL-5 cells incubated 24 h with doxorubicin (100 nM), actinomycin D (2.5 nM), and CX-5461 (2,500 nM) in the presence or absence of BCL-2. Error bars represent SD of 5 independent experiments (n = 5). Student’s t test: ∗p < 0.05, ∗∗p < 0.01. (G) Representative western blots showing the effect of 24 h treatment with doxorubicin (100 nM), actinomycin D (2.5 nM), and CX-5461 (2,500 nM) on p53 protein levels in TMD8 (BCL-2 positive) and SUDHL-5 (BCL-2 negative) cells. (H) qPCR analysis showing mRNA expression levels of TP53 and CDKN1A (p21) in TMD8 and SUDHL-5 cell lines incubated for 24 h with doxorubicin (100 nM), actinomycin D (2.5 nM), or CX-5461 (2,500 nM). Error bars represent SD of 5 independent experiments (n = 5). Student’s t test: ∗p < 0.05, ∗∗p < 0.01.

Figure 4

MDM2 inhibitors in combination with venetoclax overcome BCL-2-mediated resistance to RiBi inhibitors in vitro and in vivo (A) CTG assay showing cell viability of SUDHL-5 cells treated for 24 h with RiBi inhibitors (doxorubicin 25, 50, 100 nM; actinomycin D 0.65, 1.25, 2.5 nM; and CX-5461 625, 1250, 2,500 nM), venetoclax (125, 250, 500 nM), and with the MDM2i nutlin-3A (1,250, 2,500, 5,000 nM) as single agents or in different combinations, in the presence or absence of BCL-2. Error bars represent standard deviation (SD) of triplicate experiments (n = 3). Student’s t test: ∗p < 0.05, ∗∗p < 0.01. (B) Representative western blot showing the effects of the indicated treatments on p53 protein abundance and caspase 3 cleavage in SUDHL-5 cells in the presence or absence of BCL-2. Cells were pre-treated with doxycycline for 96 h and then incubated with doxorubicin 100 nM, actinomycin D (2.5 nM), CX-5461 (2,500 nM), venetoclax (500 nM), nutlin-3A (2,500 nM), and the indicated combinations for 24 h. (C) Bar graph depicting p53 densitometry analysis of the immunoblot shown in Figure 4B. Densitometry analyses were performed using ImageJ software. (D) CTG assay showing cell viability of TMD8 cells treated for 24 h with RiBi inhibitors (doxorubicin 25, 50, 100 nM; actinomycin D 0.65, 1.25, 2.5 nM), venetoclax (125, 250, 500 nM), and the MDM2i nutlin-3A (1,250, 2,500, 5,000 nM) as single agents or in combination. Error bars represent SD of triplicate experiments (n = 3). Student’s t test: ∗∗p < 0.01. (E) In vivo combination experiment in a subcutaneous TP53 WT/BCL-2 positive DLBCL PDX model (LNH1). NSG mice were treated with vehicle, 0.04 mg/kg actinomycin D, 50 mg/kg venetoclax, and 100 mg/kg idasanutlin (MDM2i) as single agents or in combination. Tumor volume was measured using a caliper. Error bars represent SD of five mice (n = 5). The representative western blot shows BCL-2 baseline protein level in the LNH1 PDX, as well as in SUDHL-5 and OCILY-18 cell lines (used as negative and positive controls, respectively). (F) Bar graph showing tumor volume at day 24 (same experiment Figure 4E). Error bars represent SD of five mice (n = 5). Student’s t test: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.005. (G) Representative image of three mice and their tumors at day 24 comparing vehicle-treated versus triple combination treated (actinomycin D, venetoclax, and idasanutlin).

Figure 5

BCL-2 overexpression is associated with decreased nucleolar area and adverse outcome in diffuse large B cell lymphoma (A) Dot plot graph showing BCL-2 mRNA expression levels (measured by T-GEP) in patient samples (exploratory cohort 1, n = 83) with low and high nucleolar area, as determined by quantitative image analysis of silver-stained nucleolar structures. The definition of “high” and “low” nucleolar area was based on the median value of the nucleolar area in the whole cohort. Student’s t test: ∗∗p < 0.01. (B) Dot plot graph showing values of nucleolar area in patient samples (exploratory cohort 1, n = 83) characterized by low and high BCL-2 mRNA levels. The definition of “high” and “low” BCL-2 mRNA expression was based on the median value of BCL-2 mRNA expression in the whole cohort as determined by T-GEP. Student’s t test: ∗p < 0.05. (C) Representative nucleolar silver staining of DLBCL histological sections from two patients with high (left) and low (right) BCL-2 mRNA expression. Scale bar, 10 μm. D) Dot plot graph showing the correlation significance between mRNA levels of several BCL-2 family members and nucleolar area in the exploratory cohort. Among all BCL-2 family members, only BCL-2 mRNA levels show a significantly association with nucleolar size. (E) Dot plot graph showing nucleolar area values in patient samples (validation cohort 2, n = 46) characterized by low and high BCL-2 mRNA levels. Student’s t test: ∗∗∗p < 0.005. (F) Dot plot graph showing nucleolar area values in patient samples (cohort 1 + cohort 2, n = 129) according to BCL-2 and MYC mRNA levels. L indicates “low,” and H indicates “high.” Student’s t test: ∗p < 0.05, ∗∗∗p < 0.005. (G) Progression-free survival (PFS) curve of the whole cohort (cohort 1 + cohort 2, n = 129) according to BCL-2 mRNA expression levels. p value was calculated using the log-rank test. (H) PFS curve of the whole cohort (cohort 1 + cohort 2, n= 129) according to MYC mRNA expression levels. p value was calculated using the log-rank test.

Figure 6

A reduced nucleolar area is an independent predictor of adverse outcome in DLBCL patients treated with standard anthracycline-based chemoimmunotherapy (A) Progression-free survival (PFS) curve of the whole cohort (exploratory + validation, n = 129) according to nucleolar size (nucleolar area). The definition of “high” and “low” nucleolar area was based on the median value of the whole cohort. p value was calculated using the log-rank test. (B) Overall survival (OS) curve of the whole cohort (exploratory + validation, n = 129) according to nucleolar size (nucleolar area). The definition of “high” and “low” nucleolar area was based on the median value of the whole cohort. p value was calculated with the log-rank test. (C) Forest plot depicting multivariable analyses for PFS in the whole cohort (exploratory + validation, n = 129). (D) Forest plot depicting multivariable analyses for OS in the whole cohort (exploratory + validation, n = 129).